Leveraging the Internet to Drive a Real Car in the Virtual Earth 3D Model 2024-01-2878
Digital mapping tools have become indispensable for road navigation. Applications like Waze and Google Maps harness the power of satellite imagery to provide precise visualization of GPS coordinates. The field advanced significantly in May 2023 with the introduction of dynamic 3D representations of the Earth. Companies such as Cesium now offer Unity3D and Unreal Engine Application Programming Interface that can be applied to geospatial applications. These images are no longer static and offer the opportunity to provide seamless continuous navigation.
Driving simulation has been widely used for training and research. We investigate with this project the potential of this new geospatial database as a tool for scenario development to study manual and autonomous driving. We present an in-vehicle driving simulation integration that employs a real steering wheel and pedals from a stationary vehicle as controls. The visual experience is delivered through the Meta Quest Headset through an overlay in a Mixed Reality environment.
Two case scenarios are examined. The first case involves navigating downtown Denver. The use of photorealistic representations of Denver's buildings offers an immersive experience, although the 3D topology presents some irregularities. These irregularities result from the limited number of polygons used for the digital modeling, especially on flat surfaces like roads and pavements.
The second scenario leverages the hilly landscapes outside Denver. These areas, characterized by arid, treeless terrain typical of Colorado, offer a smooth driving experience. Still, the technology incorporates projection such as phantom cars, flat images of vehicles on the roadway that were captured during satellite data acquisition. We explore opportunities to address these inaccuracies and enhance the environment for a more realistic and immersive driving experience.
